Вы находитесь на странице: 1из 12

Centers for Disease Control and Prevention

Epidemiology Program Office


Case Studies in Applied Epidemiology
No. 871-703

Screening for Antibody


to the Human Immunodeficiency Virus
Student's Guide

Learning Objectives
After completing this case study, the participant should be able to:

G Define and perform calculations of sensitivity, specificity, predictive-value positive, and


predictive-value negative;

G Describe the relationship between prevalence and predictive value;

G Discuss the trade-offs between sensitivity and specificity;

G List the principles of a good screening program.

This case study was developed in 1987 by Lyle Peterson, Guthrie Birkhead, and Richard
Dicker.

U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES


Public Health Service
CDC-EIS, 2003: Screening for HIV (871-703) – Student's Guide Page 2

PART I
In December 1982, a report in the MMWR symptoms and signs of HIV infection, those with
described three persons who had developed various autoimmune disorders and neoplastic
acquired immunodeficiency syndrome (AIDS) but diseases that could give a false-positive test
who had neither of the previously known risk result, and presumably healthy blood and
factors for the disease: homosexual/bisexual plasma donors.
activity with numerous partners and intravenous
drug use. These three persons had previously Numerous complex issues were discussed even
received whole-blood transfusions. By 1983, before licensure. Among them were
widespread recognition of the problem of understanding the magnitude of the problem of
transfusion-related AIDS led to controversial false-positive test results, and determining
recommendations that persons in known whether test-positive blood donors should be
high-risk groups voluntarily defer from donating notified.
blood. In June 1984, after the discovery of the
human immunodeficiency virus (HIV), five It is now March 2, 1985. The first HIV antibody
companies were licensed to produce test kits will arrive in blood banks in the state in a
enzyme-linked immunosorbent assay (EIA, then few hours. Meeting with State Epidemiologist to
called ELISA) test kits for detecting HIV discuss the appropriate use of this test are the
antibody. A Food and Drug Administration Commissioner of Health, the medical director of
(FDA) spokesman stated that, "...getting this test the regional blood bank, and the chief of the
out to the blood banks is our No. 1 priority...." State Drug Abuse Commission.
Blood bank directors were anxiously waiting to
start screening blood with the new test until To help in the discussions, the State
March 2, 1985, the date the first test kit was Epidemiologist turns to pre-licensure information
approved by the FDA. regarding the sensitivity and specificity of test
kit A. The information indicates that the
In the pre-licensure evaluation, sensitivity and sensitivity of test kit A is 95.0% (0.95) and the
specificity of the test kits were estimated using specificity is 98.0% (0.98). These and related
blood samples from four groups: those with measures are reviewed below.
AIDS by CDC criteria, those with other

NOTES ON SENSITIVITY AND SPECIFICITY


Actual antibody status
Test result Present Absent Total
Positive True positive (A) False positive (B) All positive tests (A+B)
Negative False negative (C) True negative (D) All negative tests (C+D)
Total All with antibody All without antibody Total (A+B+C+D)
(A+C) (B+D)

Sensitivity - the probability that the test result will Predictive-value positive (PVP) - the probability
be positive when administered to persons who that a person with a positive screening test result
actually have the antibody. actually has the antibody.
= true positives / all with antibody = true positives / all with positive test
Algebraically, sensitivity = A / (A+C) Algebraically, PVP = A / (A+B).

Specificity - the probability that the test result will Predictive-value negative (PVN) - the probability
be negative when administered to persons who that a person with a negative screening test
are actually without the antibody. result actually does not have the antibody.
= true negatives / all without antibody = true negatives / all with negative test
Algebraically, specificity = D / (B+D). Algebraically, PVN = D / (C+D).
CDC-EIS, 2003: Screening for HIV (871-703) – Student's Guide Page 3

Question 1: With this information, by constructing a 2-by-2 table, calculate the predictive-value
positive and predictive-value negative of the EIA in a hypothetical population of
1,000,000 blood donors. Using a separate 2-by-2 table, calculate PVP and PVN for a
population of 1,000 drug users. Assume that the actual prevalence of HIV antibody
among blood donors is 0.04% (0.0004) and that of intravenous drug users is 10.0%
(0.10).
CDC-EIS, 2003: Screening for HIV (871-703) – Student's Guide Page 4

The blood bank director wants assistance in antibody-positive units will be missed by the test,
evaluating the EIA as a test for screening donor and she wonders about false-positive test results
blood in the state. In particular, she is since she is under pressure to develop a
concerned about the possibility that some notification procedure for EIA-positive donors.

Question 2: Do you think that the EIA is a good screening test for the blood bank? What would you
recommend to the blood bank director about notification of EIA-positive blood donors?

The chief of the State Drug Abuse Commission survey of intravenous-drug-abuse clients and
has noticed a dramatic increase in AIDS among would like to assess the feasibility of using the
clients in his intravenous-drug-abuse treatment test results as part of behavior-modification
programs. For planning purposes, he wants to counseling.
do a voluntary HIV antibody seroprevalence

Question 3: Do you think that the EIA performs well enough to justify informing test-positive clients in
the drug abuse clinics that they are positive for HIV?
CDC-EIS, 2003: Screening for HIV (871-703) – Student's Guide Page 5

Question 4: If sensitivity and specificity remain constant, what is the relationship of prevalence to
predictive-value positive and predictive-value negative?

EIA results are recorded as optical-density (OD) with most other screening tests, is not perfect;
ratios. The OD ratio is the ratio of absorbance of there is some overlap of optical-density ratios of
the tested sample to the absorbance of a control samples that are actually antibody positive and
sample. The greater the OD ratio, the more those that are actually antibody negative. This is
"positive" is the test result. The EIA, as illustrated in the following figure.
CDC-EIS, 2003: Screening for HIV (871-703) – Student's Guide Page 6

Hypothetical distribution of results on an EIA for HIV,


by actual antibody status

Actually
without
antibody

Number of persons
Actually
have
antibody

C A B
OD Ratio

Establishing the cutoff value to define a positive intiially considered that optical density ratios
test result from a negative one is somewhat greater than "A" on the above figure would be
arbitrary. Suppose that the test manufacturer called positive.

Question 5: In terms of sensitivity and specificity, what happens if you raise the cutoff from "A" to "B"?

Question 6: In terms of sensitivity and specificity, what happens if you lower the cutoff from "A" to
"C"?

Question 7: From what you know now, what is the relationship between sensitivity and specificity of a
screening test.

Question 8: Where might the blood bank director and the head of drug treatment want the cutoff point
to be for each program? Who would probably want a lower cutoff value?
CDC-EIS, 2003: Screening for HIV (871-703) – Student's Guide Page 7

PART II
The blood bank director is concerned that, thought to be lower than that of the EIA.
because of the low predictive-value positive of Because the Western blot test is not yet
the EIA in the blood donor population, the blood generally available, the blood bank director is
bank personnel cannot properly inform those wondering whether the initial EIA-positive results
who are EIA positive of their actual antibody can be confirmed by repeating the EIA and by
status. For this reason, he wishes to evaluate considering persons to have the antibody only if
the Western blot test as a confirmatory test for results of both tests are positive.
HIV antibody.
The State Epidemiologist suggests that they
The Western blot test identifies antibodies to compare the performance of the repeat EIA and
specific proteins associated with the human the Western blot as confirmatory tests. To do
immunodeficiency virus. The Western blot is the this, they will use the earlier hypothetical sample
most widely used secondary test to detect HIV of 1,000,000 blood donors. They assume that
antibody because its specificity exceeds 99.99%; serum specimens that are initially positive by EIA
however, it is not used as a primary screening are then split into two portions; a repeat EIA is
test because it is expensive and technically performed on one portion and a Western blot on
difficult to perform. Its sensitivity is the other portion.

Question 9: What is the actual antibody prevalence in the population of persons whose blood samples
will undergo a second test?

Question 10: Calculate the predictive-value positive of the two sequences of tests: EIA-EIA and
EIA-Western blot. Assume that the sensitivity and specificity of the EIA are 95.0% and
98.0%, respectively. Assume that the sensitivity and specificity of the Western blot are
80.0% and 99.99%, respectively. Also assume that the tests are independent, even
though they may not be (e.g., those with cross-reactive proteins are likely to cross-react
each time).
CDC-EIS, 2003: Screening for HIV (871-703) – Student's Guide Page 8

Question 11: Why does the predictive-value positive increase so dramatically with the addition of a
second test? Why is the predictive value positive higher for the EIA-WB sequence than
for the EIA-EIA sequence?

It is now July 1987 and the Governor has asked positive by EIA will undergo confirmatory
the State Epidemiologist to evaluate a proposed Western blot testing.
premarital HIV-antibody-screening program. A
bill to establish the program is to be presented to The legislation describes the goal of the
the state legislature tomorrow. An estimated screening program to be to decrease inadvertent
60,000 people will get married in the state in the perinatal or sexual HIV transmission by
next year. The proposed legislation requires that determining who among those to be married are
each prospective bride and groom submit a probably infected with the virus.
blood sample for EIA testing. Samples that test

Question 12: What criteria would you consider in evaluating this proposed screening program?
CDC-EIS, 2003: Screening for HIV (871-703) – Student's Guide Page 9

The following two tables show the results of the Test Kit A available at the time were 97.0% and
testing, assuming that persons getting married 99.8%, respectively. The Western blot sensitivity
have the same actual HIV antibody prevalence and specificity were 95.0% and 99.99%,
as blood donors (0.04%). In 1987, the sensitivity respectively.
and specificity of the improved EIA

Actual antibody status


Initial EIA Present Absent Total
Positive 23 120 143 (These 143 will undergo
Western blot testing)
Negative 1 59,856 59,857

Total 24 59,976 60,000

Follow-up Western blot Present Absent Total


Positive 22 0 22
Negative 1 120 121
Total 23 120 143

With sequential tests: Sensitivity of 92%


Specificity of 100%
Predictive-value positive of 100%

Question 13: Compute the cost of the screening program. Assume a cost of $50.00 for every initial
EIA test ($10.00 lab fee and $40.00 health-care-provider visit) and an additional
$100.00 for EIA-positive persons who will need additional testing. What is the cost of
the screening program in the next year? What is the cost per identified
antibody-positive person?

Question 14: What is your final recommendation to the Governor?


CDC EIS Summer Course, 2002: Screening for HIV - Student’s Guide Page 10

THE NEW YORK TIMES NATIONAL SUNDAY, JUNE 25, 1989


Illinois Legislature Repeals Requirement for Prenuptial AIDS Tests
By ISABEL WILKERSON
Special to The New York Times

SPRINGFIELD, ILL., June 23 - At the 44 Positive Out of 221,000 But Penny Pullen, a Republican State The agency also found that the rate of
urging of health officials and AIDS Representative from suburban Cook infection among engaged couple’s was
specialists, the Illinois Legislature Since then, the tests, which detect the County, who sponsored the prenuptial comparable to those of other low-risks
repealed Friday night the only law in the antibodies that indicate infection with the AIDS testing bill, said repeal of the law groups. Engaged couples in Illinois and
country requiring prenuptial testing for human immunodeficiency virus which would hurt the state’s efforts to curb the blood donors, both groups considered at
the AIDS virus. cases AIDS, have turned up few cases of spread of the virus, “This is a major very low risk, have rates of infection of
The measure now goes to Gov. James the disease. Of the 221,000 people who mistake,” Ms.Pullen said. “This will send about 2 per 10,000.
R. Thompson. He has consistently took marriage vows in Illinois since the an unfortunate message to the people of “The overall rate among these couples
declined comment on whether he will law took effect, 44 were infected with Illinois and the rest of the nation that is close to the lowest rate ever recorded in
sign it, although pressure on him to do so the HIV virus, tests indicated, and health AIDS is not as serious an epidemic as it this country.” said Tom Schafer, a
is intense, including that of his State officials suspect that nearly a dozen of was two years ago. And that message is a spokesman for the Illinois Department of
Health Director, Dr. Bernard Turnock. those results may be false. Since the lie.” Public Health.
A similar testing law in Louisiana was testing was confidential, health officials While even critics say the law has
repealed last year, six months after it do not know the outcome of these cases. Fewer Than Predicted been useful in raising awareness of the
took effect. The tests have also led thousands of AIDS epidemic, state health officials said
“We made a mistake and we ought to people to leave the state to get married She pointed to an increase in the it was an expensive way to detect carriers
admit it,”said Bill Marovitz, a State and undetermined numbers of others to number of positive test results in the first of the virus . The test costs each person
Senator from Chicago, urging his put off marriage altogether, health half of this year as evidence that the law from $30 to $125, depending on whether
colleagues to overturn the testing law. officials said. was working. So far this year, the tests testing is done in clinics or in a doctor’s
Prenuptial testing began in Illinois in Marriages in Illinois fell by nearly a have indicated 18 cases of the AIDS office and whether follow-up testing is
January 1988 over the strong objection quarter from 99,212 in 1987 to 77,729 virus among 66,500 newly betrothed required. The total cost for Illinois couples
of both the Illinois Department of Public in 1988, although the numbers are up people, as against 8 cases among 59,000 last year was $5.4 million, or about
Health and AIDS policy experts. slightly so far this year over 1988. people in the same period last year, the $209,00 for each case of HIV infection
They said it was an inefficient and AIDS specialists hailed the repel Illinois Department of Public Health detected.
expensive way to identify carriers of the legislation as long overdue. “It’s a ‘we- said.
virus and diverted already scarce told-you-so’ situation,” said Andrew But officials of the health department
resources from those most at risk. Deppe, a spokesman for the AIDS said that even with that increase, the
Foundation of Chicago. “Illinois has agency had found far fewer case in the
become a national laughingstock. We’ve 18 months of mandatory testing than the
had to spend our energy putting out 120 cases it originally predicted would
brush fires instead of working on be found each year.
prevention.”
CDC-EIS, 2003: Screening for HIV (871-703) – Student's Guide Page 11

Appendix 1
The following 10 principles of successful mass screening programs were proposed by Wilson and Jungner
of the World Health Organization in 1968:

1. The condition being sought is an important health problem for the individual and the community.
2. There is an acceptable form of treatment for patients with recognizable disease.
3. The natural history of the condition, including its development from latent to declared disease, is
adequately understood;
4. There is a recognizable latent or early symptomatic stage.
5. There is a suitable screening test or examination for detecting the disease at the latent or early
symptomatic stage, and this test is acceptable to the population.
6. The facilities required for diagnosis and treatment of patients revealed by the screening program are
available.
7. There is an agreed policy on whom to treat as patients.
8. Treatment at the pre-symptomatic, borderline stage of a disease favorably influences its course and
prognosis.
9. The cost of the screening program (which would include the cost of diagnosis and treatment) is
economically balanced in relation to possible expenditure on medical care as a whole.
10. Case-finding is a continuing process, not a "once and for all" project.

References - Screening for HIV


1. Check WA. Preventing AIDS transmission: should blood donors be screened? JAMA 1983;
249:567-70.
2. Goldsmith MF. HTLV-III testing of donor blood imminent; complex issues remain. JAMA 1985;
253:173-81.
3. Marwick C. Use of AIDS antibody test may provide more answers. JAMA 1985; 253:1694-9.
4. Sivak SL, Wormser GP. Predictive value of a screening test for antibodies to HTLV-III. Am J Clin
Pathol 1986; 85:700-3.
5. Ward JW, Grindon AJ, Feorino PM, Schable C, Parvin M, Allen JR. Laboratory and epidemiologic
evaluation of an enzyme immunoassay for antibodies to HTLV-III. JAMA 1986; 256:357-61.
6. Cleary PD, Barry MJ, Mayer KH, et al. Compulsory premarital screening for the human
immunodeficiency virus: technical and public health considerations. JAMA 1987;258:1757-62.
7. Schwartz JS, Dans PE, Kinosian BP. Human immunodeficiency virus test evaluation, performance,
and use: proposals to make good tests better. JAMA 1988;259:2574-9.
8. Turnock BJ, Kelly CJ. Mandatory premarital testing for human immunodeficiency virus: the Illinois
experience. JAMA 1989;261:3415-8.
9. Peterson LR, White CR, and the Premarital Screening Study Group. Premarital screening for
antibodies to human immunodeficiency virus in the United States. Am J Public Health 1990;80:1087-
1090.
9. McKilip J. The effect of mandatory premarital HIV testing on marriage: the case of Illinois. Am J
Public Health 1991;81:650-3.
11. Albritton WL, Vittinghoff E, Padian NS. Human immunodeficiency virus testing for patient-based and
population-based diagnosis. J Infect Dis 1996:174(Suppl 2):S176-81.
12. Quinn TC. Acute primary HIV infection. JAMA 1997;278:58-62.
CDC-EIS, 2003: Screening for HIV (871-703) – Student's Guide Page 12

SUMMARY OF SCREENING TEST MEASURES


Condition Condition
Truly Present Truly Absent
Test positive True Positive False Positive Total Testing Positive
Test negative False Negative True Negative Total Testing Negative
Total True Prevalence 1 ! Prevalence Size of Population

Bayes Theorem Formulas for PVP and PVN: